https://bugs.kde.org/show_bug.cgi?id=521994
--- Comment #3 from Francesco Panarese <[email protected]> --- (In reply to Zamundaaa from comment #2) > Please attach the output of drm_info > (https://gitlab.freedesktop.org/emersion/drm_info), once for when the cursor > looks correct, and once for when it looks wrong. > Using the git master of drm_info would be best, since that prints more > information. [francesco@ArchLinux drm_info]$ diff ~/cursor_correct.txt ~/cursor_wrong.txt 2727,2728c2727,2728 < │ │ ├───FB ID: 426 < │ │ │ ├───Object ID: 426 --- > │ │ ├───FB ID: 424 > │ │ │ ├───Object ID: 424 2759,2760c2759,2760 < │ ├───"FB_ID" (atomic): object framebuffer = 426 < │ │ ├───Object ID: 426 --- > │ ├───"FB_ID" (atomic): object framebuffer = 424 > │ │ ├───Object ID: 424 3503c3503,3509 < │ │ ├───FB ID: 0 --- > │ │ ├───FB ID: 416 > │ │ │ ├───Object ID: 416 > │ │ │ ├───Size: 64×64 > │ │ │ ├───Format: ARGB8888 (0x34325241) > │ │ │ ├───Modifier: DRM_FORMAT_MOD_LINEAR (0x0000000000000000) > │ │ │ └───Planes: > │ │ │ └───Plane 0: offset = 0, pitch = 256 bytes 3508c3514,3520 < │ ├───"FB_ID" (atomic): object framebuffer = 0 --- > │ ├───"FB_ID" (atomic): object framebuffer = 416 > │ │ ├───Object ID: 416 > │ │ ├───Size: 64×64 > │ │ ├───Format: ARGB8888 (0x34325241) > │ │ ├───Modifier: DRM_FORMAT_MOD_LINEAR (0x0000000000000000) > │ │ └───Planes: > │ │ └───Plane 0: offset = 0, pitch = 256 bytes 3510,3512c3522,3524 < │ ├───"CRTC_ID" (atomic): object CRTC = 0 < │ ├───"CRTC_X" (atomic): srange [INT32_MIN, INT32_MAX] = 1144 < │ ├───"CRTC_Y" (atomic): srange [INT32_MIN, INT32_MAX] = 1024 --- > │ ├───"CRTC_ID" (atomic): object CRTC = 363 > │ ├───"CRTC_X" (atomic): srange [INT32_MIN, INT32_MAX] = 1515 > │ ├───"CRTC_Y" (atomic): srange [INT32_MIN, INT32_MAX] = 1214 Attached both outputs as requested. Comparing them, I noticed the cursor plane (FB 64x64, ARGB8888) is completely unused (FB_ID=0, CRTC_ID=0) when the panel is visible, and becomes active (FB_ID=416, CRTC_ID=363) once the panel auto-hides and the cursor is over the wallpaper. So the cursor only moves onto its own hardware plane once the panel disappears — which matches what I suspected from the source: with the panel always present there's always compositing, but once it's gone the wallpaper becomes the only fullscreen surface and seems to trigger direct scanout, putting the cursor on its own plane (where the color looks wrong). -- You are receiving this mail because: You are watching all bug changes.
